Scope the not-fine-tuning claim to the behavioural path

README.md

@@ -116,7 +116,7 @@ vllm serve ApolloRaines/Sharona_Q27B-R_CodeSecurity_v2 \
 
 ## About jBlaze
 
-jBlaze is a proprietary behavioral surgery tool that operates directly on model weights. It is not fine-tuning. It is not prompt engineering. It identifies behavioral directions embedded in a model's weight space and surgically modifies them to remove or amplify specific behaviors.
+jBlaze is a proprietary behavioral surgery tool that operates directly on model weights. Its behavioural edits are not fine-tuning and not prompt engineering -- they are analytic modifications of directions in weight space. Its knowledge-writing path is gradient descent on a restricted parameter set, and this checkpoint's provenance includes gradient training (see the correction above). It identifies behavioral directions embedded in a model's weight space and surgically modifies them to remove or amplify specific behaviors.
 
 jBlaze has produced 55+ models across Qwen, Llama, Gemma, Mistral, and DeepSeek architectures. Every released model is behaviorally evaluated and checked for fluency and capability degradation before release.